The Fengyun-3C Radio Occultation Sounder GNOS: a Review of the Mission and Its Early Results and Science Applications

Home , BeiDou, Radio occultation

Atmos. Meas. Tech., 11, 5797–5811, 2018 https://doi.org/10.5194/amt-11-5797-2018 © Author(s) 2018. This work is distributed under the Creative Commons Attribution 4.0 License.

The FengYun-3C radio occultation sounder GNOS: a review of the mission and its early results and science applications

Yueqiang Sun1,2,3, Weihua Bai1,2,3, Congliang Liu1,2, Yan Liu4, Qifei Du1,2,3, Xianyi Wang1,2, Guanglin Yang5, Mi Liao5, Zhongdong Yang5, Xiaoxin Zhang5, Xiangguang Meng1,2, Danyang Zhao1,2, Junming Xia1,2, Yuerong Cai1,2, and Gottfried Kirchengast6,2,1 1Beijing Key Laboratory of Space Environment Exploration, National Space Science Center, Chinese Academy of Sciences (NSSC/CAS), Beijing, China 2Joint Laboratory on Occultations for Atmosphere and Climate (JLOAC) of NSSC/CAS, Beijing, China, and University of Graz, Graz, Austria 3School of Astronomy and Space Science, University of Chinese Academy of Sciences, Beijing, China 4National Meteorological Center, Chinese Meteorological Administration, Beijing, China 5National Satellite Meteorological Center, Chinese Meteorological Administration, Beijing, China 6Wegener Center for Climate and Global Change (WEGC) and Institute for Geophysics, Astrophysics, and Meteorology/Institute of Physics, University of Graz, Graz, Austria

Correspondence: Congliang Liu ([email protected]) and Weihua Bai ([email protected])

Received: 25 October 2017 – Discussion started: 15 January 2018 Revised: 25 July 2018 – Accepted: 7 August 2018 – Published: 23 October 2018

Abstract. The Global Navigation Satellite System (GNSS) uation and scientific applications establishes confidence that Occultation Sounder (GNOS) is one of the new-generation the GNOS data from the series of FY-3 satellites will provide payloads on board the Chinese FengYun 3 (FY-3) series of important contributions to NWP, GCM, and SWR scientific operational meteorological satellites for sounding the Earth’s communities. neutral atmosphere and ionosphere. FY-3C GNOS, on board the FY-3 series C satellite launched in September 2013, was designed to acquire setting and rising radio occultation (RO) data by using GNSS signals from both the Chinese BeiDou 1 Introduction Navigation Satellite System (BDS) and the US Global Posi- tioning System (GPS). So far, the GNOS measurements and The global navigation satellite system (GNSS) radio oc- atmospheric and ionospheric data products have been vali- cultation (RO) technique (Melbourne et al., 1994; Ware dated and evaluated and then been used for atmosphere- and et al., 1996) for sounding the Earth’s neutral atmosphere ionosphere-related scientific applications. and ionosphere was demonstrated by the proof-of-concept This paper reviews the FY-3C GNOS instrument, RO data Global Positioning System/Meteorology (GPS/MET) mis- processing, data quality evaluation, and preliminary research sion launched in 1995 (Ware et al., 1996; Kursinski et al., applications according to the state-of-the-art status of the FY- 1996; Kuo et al., 1998) as well as the following GNSS RO 3C GNOS mission and related publications. The reviewed missions: the Challenging Mini-satellite Payload (CHAMP; data validation and application results demonstrate that the Wickert et al., 2001, 2002); the Constellation Observing Sys- FY-3C GNOS mission can provide accurate and precise at- tem for Meteorology, Ionosphere and Climate (COSMIC; mospheric and ionospheric GNSS (i.e., GPS and BDS) RO Anthes et al., 2000, 2008; Schreiner et al., 2007), the Gravity profiles for numerical weather prediction (NWP), global cli- Recovery and Climate Experiment (GRACE; Beyerle et al., mate monitoring (GCM), and space weather research (SWR). 2005; Wickert et al., 2005); and the Meteorological Opera- The performance of the FY-3C GNOS product quality eval- tional (MetOp) satellites (Edwards and Pawlak, 2000; Lun- tama et al., 2008). These missions have demonstrated the

Published by Copernicus Publications on behalf of the European Geosciences Union. 5798 Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS

Figure 1. Timeline of FY-3 series satellites. The FY-3 series satellites are the Chinese second-generation polar-orbiting meteorological satellites including AM (morning), PM (afternoon), and EM (evening) orbits, and rainfall types of satellites, which have been/will be launched in three batches (source: Sun et al., 2017).

unique properties of the GNSS RO technique – such as high vertical resolution, high accuracy, all-weather capability, and Figure 2. Design (functional block diagram) of GNOS instrument global coverage (Ware et al., 1996; Gorbunov et al., 1996; (source: Du et al., 2016). Rocken et al., 1997; Leroy, 1997; Steiner et al., 1999) – and long-term stability and consistency of different RO mission nals (Wang et al., 2015; Bai et al., 2018). The FY-3C satel- observations (Foelsche et al., 2009, 2011). Therefore, GNSS lite is the first BDS- and GPS-compatible RO sounder with a RO data products (i.e., bending angle, refractivity, tempera- state-of-the-art RO receiver – GNOS (Bai et al., 2012, 2014a; ture, pressure, water vapor, and ionospheric electron density Wang et al., 2015; Du et al., 2016). The following FY-3 se- profiles) have been widely used for numerical weather pre- ries of operational meteorological satellites (Fig. 1) will con- diction (NWP; e.g., Healy and Eyre, 2000; Kuo et al., 2000; tinue to carry GNOS as a major payload. The next one of Healy and Thepaut, 2006; Aparicio and Deblonde, 2008; these satellites is FY-3D, which was launched on 15 Novem- Cucurull and Derber, 2008; Poli et al., 2008; Huang et al., ber 2017 (Liao et al., 2016b; Yang et al., 2017; Sun et al., 2010; Le Marshall et al., 2010; Harnisch et al., 2013), global 2017). climate monitoring (GCM; e.g., Steiner et al., 2001, 2009, The FY-3C GNOS instrument consists of three antennas; 2011, 2013; Schmidt et al., 2005, 2008, 2010; Loescher and three radio frequency units (RFUs); and an electronic unit Kirchengast, 2008; Ho et al., 2009, 2012; Foelsche et al., (EU), which uses high-dynamic, high-sensitivity signal ac- 2011a; Lackner et al., 2011), and space weather research quisition and tracking techniques (Fig. 2). The three RFUs (SWR; Anthes, 2011; Anthes et al., 2008; Arras et al., 2008; are installed close to their corresponding antennas by us- Brahmanandam et al., 2012; Pi et al., 1997; Wickert, 2004; ing sharp cavity filters, to reduce the loss of the cable be- Yue et al., 2015). tween antennas and RFUs. The EU is the major component The development of GNSSs such as China’s BeiDou Nav- of GNOS, which accomplishes the GNSS remote-sensing igation Satellite System (BDS), Russia’s GLObal NAviga- signals’ acquisition and tracking as well as the real-time po- tion Satellite System (GLONASS), and the European Galileo sitioning and carrier phase observations (Wang et al., 2015). system has significantly enhanced the availability and capac- In FY-3C GNOS design, the different features of BDS and ity of the GPS-like satellites, which will make GNSS RO GPS signals have been taken into account, and it can observe even more attractive in the future (Bai et al., 2018). These both the neutral atmosphere and ionosphere by using both new GNSS satellites, together with planned low Earth or- BDS and GPS signals. bit (LEO) missions, will offer many more RO observations As shown in Fig. 2, the FY-3C GNOS instrument involves in the future (Wang et al., 2015; Cai et al., 2017). One of a positioning antenna, a rising occultation antenna, and a set- these LEO missions is China’s GNSS Occultation Sounder ting occultation antenna as part of its physical structure. In (GNOS) on board the FengYun 3 series C (FY-3C) satel- terms of electrical structure, the FY-3C GNOS involves five lite, which was successfully launched on 23 September 2013 antennas because each occultation antenna has an ionosphere (Wang et al., 2013, 2014; Bai et al., 2014b, 2016; Liao et al., occultation antenna and an atmosphere occultation antenna 2015, 2016a, b; Wang et al., 2016; Du et al., 2016). (Bai et al., 2014a, b). The FY-3C GNOS mission was designed and developed by the National Space Science Center, Chinese Academy of Sciences (NSSC/CAS), to sound the Earth’s neutral atmosphere and ionosphere by using both the BDS and GPS sig-

Atmos. Meas. Tech., 11, 5797–5811, 2018 www.atmos-meas-tech.net/11/5797/2018/ Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS 5799

The positioning antenna is a wide-beam, low-gain antenna the atmospheric delay (Ao et al., 2009) is also calculated on pointing toward zenith, which can track six BDS and eight board the GNOS. The baseband signal is then sampled at a GPS satellites simultaneously. Its measurements are used rate of 100 Hz. For the COSMIC and MetOp missions, the for real-time navigation, positioning, and the LEO satellites’ sampling rate of open-loop tracing is 50 Hz (Sokolovskiy, precise orbit determination (POD) in post-processing. 2001; Sokolovskiy et al., 2007, 2009) and 1000 Hz (von En- The front-view (along the satellite velocity direction) oc- geln et al., 2009), respectively. The performance of the COS- cultation antenna and back-view (satellite anti-velocity di- MIC mission proves that a 50 Hz open-loop sampling rate rection) occultation antenna are used for rising and setting is sufficient to monitor the troposphere (Sokolovskiy, 2001; occultation event tracking, respectively. The FY-3C GNOS Sokolovskiy et al., 2007, 2009), while MetOp uses 1000 Hz can track four BDS and six GPS occultation events simulta- to do a detailed spectrum analysis of the lower troposphere. neously. The atmosphere occultation antennas have a pattern Indeed, the higher the sampling rate, the more detailed the that is wide in azimuth and narrow in elevation. A gain of atmospheric information that can be obtained. Compliant approximately 10 dBi is reached over the coverage range of with the FY-3C satellite downlink capability, FY-3C GNOS about ± 35◦ in azimuth and about ±7.5◦ in elevation (Bai et adopts a 100 Hz open-loop sampling rate, which is proven al., 2014a; Du et al., 2016). to be sufficient to capture the signals modulated by the at- The EU of FY-3C GNOS is based on a field-programmable mosphere dynamics and uncertainties of the Doppler model. gate array (FPGA) and digital signal processor (DSP) frame- The designed parameters of FY-3C GNOS are summarized work. After filtering and down-conversion in the RFU, in Table 1; it can be seen that some parameters of the FY-3C the signals are digitally down-converted with an analog-to- GNOS are comparable to those of COSMIC (Rocken et al., digital converter (ADC) and then sampled at a high rate and 2000) or MetOp/GNSS Receiver for Atmospheric Sounding transmitted to the channel processor of the EU, where the (GRAS; Loiselet et al., 2000). GNOS accomplishes navigation, positioning and occulting After this introduction to the topic and the GNOS instru- GNSS satellite prediction and selection, signal acquisition ment, the paper is structured as follows. Section 2 provides a and tracking, and data handling. An ultra-stable oscillator description of FY-3C GNOS data processing and data prod- (USO) is used as a reference oscillator with highly stable fre- ucts. Section 3 describes the validation and evaluation of FY- quency (1 s Allan deviation of 10−12), in order to retrieve 3C GNOS data. Section 4 presents the GNOS RO data ap- atmospheric measurements with high accuracy (Du et al., plications. Finally, a summary and conclusions are given in 2016; Sun et al., 2017). It also allows data users to invert the Sect. 5. excess phase by using the zero-difference method (Beyerle et al., 2005; Bai et al., 2018). The FY-3C GNOS is a multi-frequency receiver with 2 FY-3C GNOS data processing and data products BDS and GPS compatibility, BDS B1 and B2 closed-loop The FY-3C satellite flies in a sun-synchronous polar orbit tracking, GPS L2 codeless-mode operation for P code, GPS of inclination 98.8◦, mean altitude 836 km, and orbital pe- L2C closed-loop tracking, and GPS L1 C/A closed-loop and riod 101.5 min. Therefore, GNOS can observe the tropo- open-loop tracking capabilities. The capability for both BDS sphere, stratosphere, and ionosphere from the Earth surface and GPS increases the number of transmitting sources and to around 800 km altitude. As a multi-GNSS receiver, GNOS promises significant enhancements in total throughput of has the capability of observing the phases and amplitudes of measurements. The FY-3C GNOS receiver measures the fol- radio waves from GPS and BDS satellites as they are oc- lowing observable parameters: for each tracked GPS satel- culted by the Earth’s atmosphere. From the raw GNOS ob- lite, L1 C/A code phase, L1 carrier phase, L1 signal am- servations, accurate and precise vertical bending angle pro- plitude, L2 P code phase, L2C code phase (if present), L2 files are obtained in the troposphere, stratosphere, and iono- carrier phase, and L2 signal amplitude; and for each tracked sphere. Based on the bending angles, profiles of atmospheric BDS satellite, B1I code phase, B1 carrier phase, B1 signal and ionospheric refractivity are calculated. The refractivity amplitude, B2I code phase, B2 carrier phase, and B2 signal is a function of temperature, pressure, water vapor pressure, amplitude (Bai et al., 2014a). and electron density, so the refractivity profiles can be used In the lower part of the troposphere, where highly dy- to derive profiles of temperature and water vapor in the tropo- namic signal conditions are frequently encountered due sphere, temperature in the stratosphere, and electron density to the strong atmospheric modulations, the GPS L1 sig- in the ionosphere. The operational processing procedure and nal is tracked in open loop in parallel with the closed- the product levels of the FY-3C GNOS RO data are shown in loop tracking. In open-loop tracking, the signal is down- Fig. 3. converted using a numerically controlled oscillator, which generates a frequency given by an onboard Doppler model pre-calculated in GNOS without feedback from the received signal (Sokolovskiy, 2001; Sokolovskiy et al., 2009). Partic- ularly, for the rising occultation, an a priori range model of www.atmos-meas-tech.net/11/5797/2018/ Atmos. Meas. Tech., 11, 5797–5811, 2018 5800 Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS

Table 1. Main parameters of the FY-3C GNOS instrument (source: Bai et al., 2014; Liao et al., 2016b).

Parameters Content GNSS signals GPS L1, L2; BDS B1, B2 Channel numbers Positioning: 8; occultation: 6 (GPS), 4 (BDS) Sampling rate Positioning & ionosphere occultation: 1 Hz Atmosphere occultation: closed loop 50 Hz, open loop 100 Hz Output observations Type: L1C/A, L2C, L2P/ B1I, B2I Contents: pseudo-range/carrier phase/SNR Clock stability 1 × 10−12 (1 s Allan) The peak gain of occultation antenna 10 dBi (azimuth: ±35◦; horizontal: ±7.5◦) Pseudo-range precision ≤ 30 cm Carrier phase precision ≤ 2 mm Real-time positioning precision ≤ 10 m Total weight ≤ 14 kg Total power ≤ 40 W Equivalent noise temperature 250 K Signal bandwidth GPS P code: 20.46 MHz BDS: 4.092 MHz

2.1 FY-3C GNOS mission system architecture ing center in Beijing. At the data processing center, firstly the raw observed data are decrypted and decompressed, and As shown in Fig. 3, the FY-3C GNOS mission system con- then the data packets are classified and stored in 12 different sists of three major segments, i.e., GNSS satellite segment, specific storage spaces, according to the number of different LEO satellite segment, and ground segment. The GNSS seg- instrument payloads. One of the 12 data packets is the binary ment is composed of the GPS system and the (currently still format FY-3C GNOS observations, mainly including phase regional) BDS system. The latter includes five geostation- and signal-to-noise ratio (SNR) measurements, and these raw ary orbit (GEO) satellites, five inclined geosynchronous or- data are defined as the Level 0 data. bit (IGSO) satellites, and four medium Earth orbit (MEO) orbit satellites available to conduct radio occultation. The LEO satellite segment is composed of the FY-3C satellite 2.3 FY-3C GNOS Level 1 data processing carrying the GNOS RO receiver. The ground segment consists of a data processing center in Beijing and five ground stations, located in Beijing, Wulumuqi, Guangzhou, Jiamusi, The unpacking and interpretation program module verifies and Kiruna. The ground stations are mainly used to receive and revises the data package format of the FY-3C Level 0 observed data from the FY-3C satellite and then transmit raw data and then unpacks them into seven packages, i.e., them to the data processing center. In addition, in the ground GPS positioning package, GPS ionospheric occultation pack- segment, auxiliary information provided by the international age, GPS atmospheric closed-loop occultation package, GPS GNSS service (IGS) stations – such as the GPS/BDS precise atmospheric open-loop occultation package, BDS position- orbits, clock files, Earth orientation parameters, and the co- ing package, BDS ionospheric occultation package, and BDS ordinates and measurements of the ground stations – are also atmosphere occultation package; finally the program pro- needed. cesses and stores the data according to the naming rules and data storage principles. The format conversion program mod- 2.2 FY-3C GNOS Level 0 data preparation ule subsequently converts the data into Receiver-Independent Exchange Format (RINEX) or Network Common Data Form In total there are 12 instrument payloads on board the FY- (NetCDF) format. According to the time period of the ob- 3C satellite, and their observed data are downloaded to the served data, the auxiliary data acquisition program module ground data stations and then transferred to the data process- automatically downloads the BDS and GPS ephemeris and

Atmos. Meas. Tech., 11, 5797–5811, 2018 www.atmos-meas-tech.net/11/5797/2018/ Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS 5801

Based on the LEO POD data and occultation observations, the excess-phase calculation program module is run to determine excess-phase data. Currently, in the processing of GNOS data, the single-differencing method is applied to obtain the excess phase as a function of time in an Earth- centered inertial reference frame, for all the GPS RO events and the BDS RO events which have sufﬁcient reference satellites. Under the condition of fewer reference satellites, a zero-difference technique is applied and more appropriate for BDS RO events, since it does not require a reference satellite for simultaneous observations but requires an ultra-stable oscillator on an LEO receiver (Beyerle et al., 2005), which is available for GNOS (cf. Table 1; Bai et al., 2018). Based on the encouraging results of Bai et al. (2018), an in-depth comparison of the zero-differencing and single-differencing FY-3C GNOS RO data products has been started to evaluate whether the zero-differencing method can be used instead of the single-difference method in the operational data processing.

2.4 FY-3C GNOS Level 2 data processing

The data pre-processor module includes the atmospheric and ionospheric occultation data pre-processing modules, which mainly involve quality control, data filtering, and open-loop data quality control and initial processing. The atmosphere inversion module includes the atmospheric impact parameter and bending angle calculation; inversion to refractivity; and temperature, pressure, and hu- Figure 3. Overview diagram of the elements of the FY-3C GNOS midity retrieval. The Radio Occultation Processing Package mission and its data processing and data product levels, from raw (ROPP) software (V6.0) developed at the Radio Occultation data (Level 0) down to retrieved atmospheric and ionospheric profiles (Level 2). Meteorology Satellite Application Facility (ROM SAF) is used for this purpose (Offiler, 2008; Culverwell et al., 2015). More specifically, from the excess phase the Doppler fre- almanac data, from the IGS website and other specific web- quency can be obtained; then the bending angle as a func- sites, for use by the LEO POD module. tion of impact parameter can be determined from the Doppler Highly accurate measurements by the GNSS and LEO frequency shift and the corresponding satellite positions and satellites in terms of time and position are the key to suc- velocities. Two distinct approaches have been used to de- cessful retrieval of the atmospheric and ionospheric profiles termine bending angle profiles: above an altitude of 25 km, of an occultation event. According to the GNSS position- where single-path rays propagate through the atmosphere, ing package data and satellite precision ephemeris data, the the geometric optics (GO) algorithm is used (Kursinski et POD program module conducts the LEO POD to obtain LEO al., 1997), while below 25 km, where there can be multi- satellite accurate position, velocity, satellite attitude, receiver path effects, the wave optics (WO) algorithm of Gorbunov clock error, and clock drift information. Based on the mea- et al. (2004, 2011) is used, also referred to as the canonical surements of pseudo-range and carrier phase as well as the transform (CT2)) algorithm. In order to retrieve neutral atmo- attitude information of the GNOS POD antenna, the GNSS spheric parameters, the ionosphere effects on the bending an- clock offsets, GNSS precise orbit information, and the Earth gles need to be eliminated. For the GNSS L band signals, the orientation parameters, the LEO POD can be conducted by ionosphere refractivity is proportional to the inverse square integrating the equations of celestial motion. Currently, in of the frequencies, whereas the neutral atmosphere refractiv- FY-3C GNOS data processing, the Bernese software (V5.0) ity is almost independent of the frequencies (Vorob’ev and (Dach et al., 2007) and the Position And Navigation Data An- Krasil’nikova, 1994; Syndergaard et al., 2000). Therefore, alyst (PANDA) software (Zhao et al., 2017) have been used a dual-frequency linear combination can mostly correct the and evaluated, and both of them can deliver highly accurate first-order ionospheric effects (Vorob’ev and Krasil’nikova, POD results. 1994). However, there are some higher-order ionospheric effects that still remain in the bending angle profiles (Kursinski www.atmos-meas-tech.net/11/5797/2018/ Atmos. Meas. Tech., 11, 5797–5811, 2018 5802 Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS et al., 1997; Liu et al., 2013, 2015, 2016, 2018). To reduce frequency difference signal (e.g., Syndergaard et al., 2000), the ionospheric residual errors and other small-scale noise, 2 2 the statistical optimization technique is used together with f f TEC = 1 2 (L − L ), (4) the MSISE-90 climatology model. An optimal linear combi- 2 − 2 1 2 C f1 f2 nation is expressed as a matrix equation to compute the neutral atmospheric bending angle and the ionospheric bending where L1 and L2 are the dual-frequency carrier phase obser- 3 −2 angle (Gorbunov, 2002). vations, the constant C = 40.3082 m s , and f1 and f2 are After ionospheric correction, under the assumption of lo- the two signal frequencies. Then the electron density Ne in cal spherical symmetry, the refractive index can be retrieved the ionosphere is derived by using an inverse Abel transfor- by an Abel transform from a given bending angle profile α mation, as a function of impact parameter a, as shown in Eq. (1) LEO (Fjeldbo et al., 1971; Melbourne et al., 1994; Kursinski et 1 Z dTEC/da = − al., 1997), and then refractivity N can be obtained from the Ne(a0) q da, (5) π a2 − a2 refractive index n as shown in Eq. (2): a0 0

 ∞  where Ne is electron density and a the impact parameter for 1 Z α(a) = − ionospheric altitudes. We note that the orbit of the FY-3C n(a0) exp q da, (1) π 2 2 satellite is at 833 km height, and our simulation estimation a a − a 0 0 indicated that the ionosphere above the LEO satellite affects the accuracy of the retrieved results slightly, but the relative bias was less than 0.1 %. Therefore, the ionosphere above the 6 N = (n − 1) × 10 . (2) LEO satellite can be neglected in the FY-3C GNOS situation. A simulation study (Wu et al., 2009) showed that the horizon- The refractivity is a function of temperature (T ), pressure tal gradients could bring ∼ 6 % to ∼ 10 % root-mean-square (p), water vapor pressure (e), and electron density (ne), as error (RMSE), which is comparatively much higher. shown in Eq. (3). p e n N = 77.6 + 3.73 × 105 − 4.03 × 107 e , (3) 3 Preliminary validation and evaluation of the FY-3C T T 2 f 2 GNOS data where f is the frequency of the GNSS signal. Therefore the 3.1 Evaluation of the FY-3C satellite POD data refractivity profiles can be used to derive profiles of temperature, water vapor, and ionospheric electron density. The FY-3C GNOS observations and its POD results (e.g., Because of the ambiguity of temperature and humidity in clock estimates, position, and velocity) are fundamental and the lower troposphere (Healy and Eyre, 2000; Poli et al., crucial elements of the whole GNOS RO data processing 2002), one-dimensional variation (1-D-Var) analysis, involv- chain. Thus, the FY-3C satellite POD performance has been ing collocated profiles of the Chinese global forecast model evaluated by difference GNSS data processing centers (Liao (used as background at T639L60 resolution), is used to re- et al., 2016b; Li et al., 2017; Xiong et al., 2017; Zhao et al., trieve temperature and humidity profiles (Liao et al., 2016b). 2017) by using the Bernese software (Liao et al., 2016b) and The ionosphere inversion module involves the ionospheric PANDA software (Zhao et al., 2017) separately. In all these total electron content (TEC) calculation and the electron den- POD data quality evaluations, an internal consistency metrics sity retrieval, in which the dual-frequency difference method method named overlapping orbit differences (OODs) (Zhao is used. The FY-3C GNOS ionospheric occultation data et al., 2017) has been used. mainly include dual-frequency carrier phase and SNR ob- The Li et al. (2017) study showed the overlapping orbit servations, with a sampling rate of 1 Hz. Since the primary consistency, and the 3-D RMSE of OODs was found to be mission of FY-3C GNOS is the neutral atmosphere occulta- 2.7 cm by using GPS data only and 3.4 cm by using both tion sounding, when there is no free channel, a new atmo- GPS and BDS data together. Furthermore, the 3-D RMSE spheric occultation event will take over and occupy an iono- of OODs was found to be 30.1 cm by using all the BDS data spheric occultation channel. Therefore, the number of com- and 8.4 cm by using the MEO and IGSO BDS data only. Sim- pleted ionospheric occultation profiles is less than that of the ilarly, the Xiong et al. (2017) study also showed overlapping atmospheric RO events, and there are typically around 220 orbit consistency, and the 3-D RMSE of OODs was found to GPS and 130 BDS ionospheric RO events per day. be about 3.8 cm by using both GPS and BDS data together, When GNSS signals propagate through the ionosphere and about 22 cm by using BDS data only. from GNSS satellites to the FY-3C satellite, their propaga- Zhao et al. (2017) used the GPS and BDS observed data tion paths are bent and delayed by ionosphere refraction ef- to conduct the POD of the FY-3C satellite. The solutions fects. The TEC can hence be calculated by using the dual- showed that the 3-D RMSE of 6 h OODs reached 2.3 and

Atmos. Meas. Tech., 11, 5797–5811, 2018 www.atmos-meas-tech.net/11/5797/2018/ Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS 5803

Figure 4. Distribution of the FY3 C-GNOS radio occultation events on 2 October 2013. There were 94 rising BDS RO events (upward- pointing red triangle), 90 setting BDS RO events (downward-pointing red triangle), 287 rising GPS RO events (upward-pointing blue triangle), and 256 setting GPS RO events (downward-pointing blue triangle) (source: Wang et al., 2015).

Table 2. The overall FY-3C POD performance. the zero-differencing algorithm with a least-squares estima- tor method and dual-frequency ionospheric correction has Parameters Content been conducted (Cai et al., 2017). The quality analysis of the FY-3C POD was conducted by using 2 months of data. Real-time navigation error (RMSE) radial 6.74 m tangential 4.20 m The results showed that the mean RMSE of 6 h OODs along normal 3.00 m radial, tangential, and normal directions is 1.24, 1.60, and 3.07 cm, respectively (Cai et al., 2017). The overall FY-3C POD error (RMSE) radial 1.24 cm POD performance – including velocity real-time position and tangential 1.60 cm velocity errors, POS position and velocity errors, and clock normal 3.07 cm stability (100s Allan deviation) – is summarized in Table 2. Clock stability (1 s Allan deviation) 10−12

3.2 Validation and evaluation of the FY3 C-GNOS atmospheric profiles 15.8 cm for GPS-only and BDS-only procedures, respectively. The quality of the FY-3C POD result calculated by us- After the launching of the FY-3C satellite, an in-orbit testing ing the BDS data is worse than that calculated by using GPS of the FY-3C GNOS data was presented by several papers data mainly because of the limited number of BDS satel- (Wang et al., 2015; Liao et al., 2017; Bai et al., 2018). lites and the lower quality of BDS satellite POD data due In the Wang et al. (2015) study the GNOS RO events and to the small number of ground stations as well as restricted their global distribution were analyzed; comparing with the distribution of ground stations. In the same study, Zhao et GPS RO observations, the accuracy and consistency of BDS al. (2017) improved the quality of BDS satellite POD data real-time positioning results and BDS RO products were an- by using ground station data and accurate FY-3C POD data, alyzed. The preliminary results showed that, compared with which were retrieved from GPS observations. the number of GPS GNOS RO events, the regional, incom- At NSSC, both the Bernese and PANDA software have plete BDS system with 14 navigation satellite can enhance been used in the FY-3C satellite POD processing, and in the number of RO events by about 33 % (Fig. 4). The statisti- this paragraph the POD performance of PANDA software cal BDS and GPS GNOS RO data analyses, by using 17 pairs is briefly discussed. Since the BDS is an incomplete con- of BDS and GPS GNOS RO events in a week, showed that stellation and can only provide regional navigation and po- the BDS–GPS difference in standard deviation of refractiv- sitioning services, mainly in the Asia-Pacific area, the FY- ity, temperature, humidity, pressure, and ionospheric electron 3C GNOS satellite POD processing is implemented by using density is lower than 2 %, 2 K, 1.5 g kg−1, 2 %, and 15.6 %, the reduced-dynamic orbit determination method in which respectively. Therefore, the BDS observations/products are www.atmos-meas-tech.net/11/5797/2018/ Atmos. Meas. Tech., 11, 5797–5811, 2018 5804 Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS

Figure 5. Comparisons of GNOS GPS, COSMIC, and Figure 6. Refractivity deviation from the ECMWF analysis for FY- MetOp/GRAS in terms of raw bending angles (from 1 November to 3C GNOS GPS (from 1 November to 31 December 2013). The left 31 December 2013). In (a), the solid curves show the mean biases panel shows the mean bias (black) and the standard deviation (red), for the two collocation ensembles, and the dashed curves show and the right panel shows the samples used vs. altitude (source: Liao the standard deviations. (b) shows the collocation ensemble size, et al., 2016b). increasing with altitude since less collocated proﬁles reach into the lower troposphere (source: Liao et al., 2016b).

in general consistent with those from GPS (Wang et al., 2015). Taking COSMIC and MetOp/GRAS RO data as bench- marks, the GNOS GPS raw bending angles (i.e., bending angles after combining L1 and L2 frequencies but before the process of statistical optimization) were evaluated. The coincident RO events of the different missions are defined to collocate in time within 3 h and in distance within 200 km (Liao et al., 2016b). The distance is defined as the distance of tangent point locations of the collocated pair of occultations Figure 7. Refractivity deviation from the ECMWF analysis for FY- at a height of 30 km (this means that the distance of some 3C GNOS BDS. The description is the same as for Fig. 6 (source: point pairs in the RO profiles may be larger than 200 km). Liao et al., 2016b). For the period from 1 November to 31 December 2013, there were 17 509 GNOS GPS, 32 588 MetOp, and 22 821 COS- MIC occultations to build 1654 GNOS–MetOp coincident the mean bias of the refractivity obtained through GPS (BDS) pairs and 2886 GNOS–COSMIC pairs. We did not yet in- GNOS was approximately −0.09 % (−0.04 %) from the near clude the data of GNOS BDS for which it could be expected surface to about 45 km. While the average standard deviation that GPS and BDS show similar features, but it will be inter- was approximately 1.81 % (1.26 %), it was as low as 0.75 % esting and hence undertaken to probe the details of GPS and (0.53 %) in the range of 5–25 km, where best sounding re- BDS differences in future work. sults are usually achieved. Figure 5 shows the mean and standard deviations of the Since FY-3C GNOS uses an ultra-stable oscillator with 1 s bending angles in terms of percentage for the two colloca- stability (Allan deviation) at the level of 10−12, both zero- tion ensembles. The features of standard deviation changing differencing and single-differencing excess-phase processing with respect to altitude are similar to the mean bias. The methods are basically feasible for FY-3C GNOS observa- best fit between GNOS, COSMIC, and MetOp, in terms of tions. Focusing on the evaluation of the bending angle and mean bias, occurs over the vertical range of 7–25 km. Some refractivity profiles that derived from zero-differencing and relatively obvious positive and negative deviations exist be- single-differencing excess-phase data of BDS RO, a compar- low 7 km and above 25 km, respectively, where the SNRs ison analysis has been conducted by using a 3-month set of are lower. For detailed discussions, please refer to Liao et GNOS BDS RO data (October to December 2013) and the al. (2016b). collocated profiles from ECMWF analyses (Bai et al., 2018). Taking the collocated European Centre for Medium-Range The statistics showed that the results from single and zero Weather Forecasts (ECMWF) analysis model data as refer- differencing are consistent in both bias and standard devia- ence (“true”) values, the quality of atmospheric refractivity tion, and also demonstrated the feasibility of zero differenc- profiles from the FY-3C GNOS mission have been evaluated ing for GNOS BDS RO observations. The average bias (and by Liao et al. (2016b). The results (Figs. 6 and 7) showed that standard deviation) of the bending angle and refractivity pro-

Atmos. Meas. Tech., 11, 5797–5811, 2018 www.atmos-meas-tech.net/11/5797/2018/ Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS 5805

Figure 8. Comparison of NmF2 measurements from GNOS GPS (a) and BDS (b) occultation and ionosonde data. The linear regression of the absolute NmF2 values was computed using a standard difference technique, in which the black line is y = x, the red line is the ﬁtted regression, corr. coef. is the correlation coefﬁcient, bias is the statistical bias, and SD is the standard deviation (source: Bai et al., 2018).

ﬁles was found to be as small as about 0.05 % to 0.2 % (and Table 3. The statistics of the GNOS GPS and BDS RO NmF2 index. 0.7 % to 1.6 %) over the upper troposphere and lower stratosphere, including for the GEO, IGSO, and MEO subsets (Bai Parameters Content et al., 2018). GPS Nmf2 bias 3.00 % GPS Nmf2 SD 17.93 % 3.3 Validation and evaluation of the FY3 C-GNOS BDS Nmf2 bias 4.67 % ionospheric proﬁles BDS Nmf2 SD 19.19 %

We evaluated the FY-3C GNOS ionosphere electron density profiles through a statistical comparison analysis with ionosonde data, in which both the BDS and GPS GNOS RO 4 Applications of FY-3C GNOS data products data were compared against observations from 69 ionosonde stations. For detailed information of FY-3C GNOS iono- 4.1 Applications of the atmospheric products sphere occultation data processing, and the evaluation analysis algorithms, we refer to Yang et al. (2017). Relevant sta- The FY-3C GNOS data, mainly including bending an- tistical results are shown in Fig. 8 and Table 3. The linear gle and refractivity profiles, have been assimilated in regression of absolute maximum electron density in F2 layer the Global/Regional Assimilation and Prediction System (NmF2) values derived from the GNOS GPS occultation and (GRAPES) of the China Meteorological Administration ionosonde data gives a correlation coefficient of 0.95, sta- (CMA; Liu et al., 2014) for the NWP application. Figure 9 tistical bias of 3.0 %, and standard deviation of 17.9 %, in shows the effects of the GPS GNOS RO data only (red line) which a total of 547 matching pairs of data were used. Sim- and both the BDS and GPS GNOS RO data (blue line), used ilarly, the linear regression of absolute NmF2 values derived in 7-day forecast at 500 hPa altitude level, on the NWP re- from the GNOS BDS occultation and ionosonde data gives a sults, in which the black line denotes the reference line, and correlation coefficient of the fitted regression of 0.95, statisti- the x axis and y axis denote the forecast time range and cal bias of 4.7 %, and standard deviation of 19.2 %, in which the anomaly correlation coefficient (ACC), respectively. The a total of 376 matching pairs of data were used. One can ACC can be regarded as a skill score relative to the climate see that the bias and standard deviation of the NmF2 derived knowledge and is positively orientated, with increasing ACC from the GNOS BDS occultation and GPS occultation are values indicating increasing “success”. The results indicated consistent and comparable (Yang et al., 2017). The statistics that FY-3C GNOS RO data have a positive effect on analysis of GNOS NmF2 are also in line with the CHAMP mission, and forecast in all medium-term forecast ranges in GRAPES, whose NmF2 average bias is −1.7 % and standard deviation not only in the Southern Hemisphere, where conventional ob- is 17.8 % (Jakowski et al., 2002). servations are lacking, but also in the Northern Hemisphere, where data are more dense. Figure 10 shows an evaluation score card of the effects of the FY-3C GNOS RO data on the GRAPES forecast results, in which the grey color denotes that the effect is tiny, the red color denotes positive effect, and the green color denotes negative effect. The evaluation results showed, that except for www.atmos-meas-tech.net/11/5797/2018/ Atmos. Meas. Tech., 11, 5797–5811, 2018 5806 Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS

Figure 9. Comparison of the anomaly correlation coefﬁcients of the without-GNOS-RO-data reference case (black line) and the other cases, assimilated GPS GNOS RO data only (red line), and assimilated GPS and BDS GNOS RO data (blue line). The left and right panels show the results of the Northern and Southern Hemisphere, respectively.

Figure 10. Evaluation score card of GPS and BDS FY-3C GNOS RO data assimilation effects on the GRAPES forecast model.

Atmos. Meas. Tech., 11, 5797–5811, 2018 www.atmos-meas-tech.net/11/5797/2018/ Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS 5807 somewhere in the Northern Hemisphere and East Asia, the GPS and BDS GNOS data tend to have overall clearly positive effects on the GRAPES NWP results. Since June 2017, the GNOS RO data are therefore operationally assimilated into GRAPES by the CMA.

4.2 Applications of the FY3 C-GNOS ionospheric products

The FY-3C GNOS ionospheric products have been applied in the ionospheric and space weather research areas. A comparison has been made between ionospheric peak parameters retrieved by FY-3C GNOS RO data and those measured by globally distributed ionosondes (Mao et al., 2016). Reason- able agreement was obtained in this case; the results indicated that NmF2 and hmF2 (the height of the F2 maximum) retrieved from FY-3C GNOS measurements are reliable and can be used for ionospheric physics studies. The comparison between the FY-3C GNOS data and the International Refer- ence Ionosphere (IRI) model is also reasonably good, but the IRI model tends to overestimate NmF2 at the crests of the equatorial anomalies (Mao et al., 2016). In addition, Yang Figure 11. Comparison of average NmF2 values from 17 ionosonde et al. (2017) analyzed sporadic E-layer events, caused by stations and GNOS ionosphere data in the region of magnetic incli- ◦ precipitating particle in the auroral region, by using FY-3C nations between 40 and 80 in the Northern Hemisphere (source: GNOS observations. The results showed that the disturbance Bai et al., 2016). intensity of the sporadic E layers in the summer hemisphere is significantly greater than that in the same latitude region in the winter hemisphere. Additionally, the occurrence rate of sporadic E layers in the summer hemisphere was found to be and the results showed that FY-3C satellite POD accuracy significantly higher than that in the winter hemisphere. reaches centimeter level. Based on the validated FY-3C GNOS ionospheric data and Regarding the GNOS atmospheric data, compared with ionosonde stations observations, the global ionospheric ef- collocated ECMWF analysis data, the mean bias (standard fects of the strong magnetic storm event in March 2015 were deviation) of the atmospheric refractivity is less than 1 % analyzed by Bai et al. (2016). Both the analyses of GNOS (2 %) in the upper troposphere and lower stratosphere. When ionospheric data and ionosonde station observations showed the GNOS ionospheric data are compared with ionosonde that the magnetic storm caused a significant disturbance in station data, the statistical bias (standard deviation) of both NmF2 and hmF2 levels, and the average NmF2 featured the the BDS and GPS GNOS ionospheric electron density NmF2 same basic trends in the zone of geomagnetic inclination be- values is less than 5 % (20 %). The consistency of BDS and ◦ tween 40 and 80 . Suppressed daytime and nighttime NmF2 GPS RO products has been successfully validated as well; levels indicated mainly negative storm conditions (Fig. 11). it has proven the utility of the GNOS RO data for applica- The analysis in this way also demonstrated the value of the tions in numerical weather prediction (NWP), global climate FY-3C GNOS data and especially confirmed the utility of its monitoring (GCM), and space weather research (SWR). ionosphere products for statistical and event-specific iono- So far the GNOS RO data products have been widely used spheric physics analyses. in NWP and SWR applications in China mainly. In the near future, the GNOS RO data will also be heavily used for GCM applications, through cooperation research between NSSC 5 Summary and conclusions (Beijing, China) and WEGC (Graz, Austria). Wider international use in NWP and other applications will also follow FY-3C GNOS is the first BDS- and GPS-capable GNSS RO based on the continuously improving data product quality of sounder in space and combines a state-of-the-art RO receiver, the GNOS data. which has meanwhile been successfully running in orbit for With the further expansion of the GNSS transmitter satel- more than 4 years. So far, a large data set of FY-3C GNOS lite constellations, and the additional GNOS instruments RO observations has been obtained, which includes both launched on board the FY-3 series of satellites, the FY-3 atmospheric and ionospheric RO products. These products GNOS observations are expected to provide an essential fu- have been validated and evaluated by difference institutions, ture contribution to the pool of international high-quality at- www.atmos-meas-tech.net/11/5797/2018/ Atmos. Meas. Tech., 11, 5797–5811, 2018 5808 Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS mospheric and ionospheric RO products, for substantial ben- Aparicio, J. and Deblonde, G.: Impact of the assimilation of efit of NWP, GCM, and SWR. CHAMP refractivity profiles in Environment Canada global forecasts, Mon. Weather Rev., 136, 257–275, 2008. Arras, C., Wickert, J., Beyerle, G., Heise, S., Schmidt, T., and Ja- Data availability. The ECMWF (Reading, UK) provides access to cobi, C.: A global climatology of ionospheric irregularities de- their archived analysis and forecast data, available at http://www. rived from GPS radio occultation, Geophys. Res. Lett., 35, 137– ecmwf.int/en/forecasts/datasets (last access: 23 August 2018), and 149, https://doi.org/10.1029/2008gl034158, 2008. NOAA-NCEI (Boulder, CO, USA) provides access to their In- Bai, W. H., Sun, Y. Q., Du, Q. F., and Wang, X. Y.: FY3 – GNOS tegrated Global Radiosonde Archive (IGRA), available at http:// Instrument Performance and Results Analysis in Mountain-based www.ncdc.noaa.gov/data-access/weather-balloon-data (last access: Validation Experiment, COSPAR 2012, India, 2012. 23 August 2018). GNOS-related data on the results of this study are Bai, W. H., Sun, Y. Q., Du, Q. F., Yang, G. L., Yang, Z. D., Zhang, currently not in the public domain and therefore cannot be accessed P., Bi, Y. M., Wang, X. Y., Cheng, C., and Han, Y.: An intro- online; in case of specific interests, data requests can be sent to the duction to the FY3 GNOS instrument and mountain-top tests, corresponding authors. Atmos. Meas. Tech., 7, 1817–1823, https://doi.org/10.5194/amt- 7-1817-2014, 2014a. Bai, W. H., Sun, Y. Q., Du, Q. F., Yang, G. l., Yang, Z. D., Zhang, P., Bi, Y. M., Wang, X. Y., Wang, D. W., and Meng, X. G.: An Competing interests. The authors declare that they have no conflict introduction to FY3 GNOS in-orbit performance and preliminary of interest. validation results, EGU General Assembly, Vienna, Geophysical Research Abstracts, 16, EGU2014-4036, 2014b. Bai, W., Wang, G., Sun, Y., Shi, J., Meng, X., Wang, D., Du, Acknowledgements. This research was supported by the National Q., Wang, X., Xia, J., Cai, Y., Liu, C., Li, W., Wu, C., Natural Science Foundation of China (grant nos. 41775034, Zhao, D., Wu, D., and Liu, C.: Application of Fengyun 3-C 41405039, 41405040, 41505030, and 41606206), the Strategic GNSS occulation sounder for assessing global ionospheric re- Priority Research Program of the Chinese Academy of Sciences sponse to magnetic storm event, Atmos. Meas. Tech. Discuss., (grant no. XDA15012300), the Scientific Research Project of the https://doi.org/10.5194/amt-2016-291, in review, 2016. Chinese Academy of Sciences (grant no. YZ201129), and the Bai, W., Liu, C., Meng, X., Sun, Y., Kirchengast, G., Du, Q., Wang, FengYun 3 (FY-3) Global Navigation Satellite System Occultation X., Yang, G., Liao, M., Yang, Z., Zhao, D., Xia, J., Cai, Y., Liu, Sounder (GNOS) development and manufacture project led by L., and Wang, D.: Evaluation of atmospheric profiles derived NSSC/CAS. The research at WEGC was supported by the Austrian from single- and zero-difference excess phase processing of Bei- Aeronautics and Space Agency of the Austrian Research Promotion Dou radio occultation data from the FY-3C GNOS mission, At- Agency (FFG-ALR) under projects OPSCLIMVALUE (grant no. mos. Meas. Tech., 11, 819–833, https://doi.org/10.5194/amt-11- 848013) and EOPCLIMTRACK (grant no. 859773). 819-2018, 2018. Beyerle, G., Schmidt, T., Michalak, G., Heise, S., Wickert, J., and Edited by: Joanna Joiner Reigber, C.: GPS radio occultation with GRACE: Atmospheric Reviewed by: three anonymous referees profiling utilizing the zero difference technique, Geophys. Res. Lett., 32, L13806, https://doi.org/10.1029/2005GL023109, 2005. Brahmanandam, P. S., Uma, G., Liu, J. Y., Chu, Y. H., Latha Devi, N. S. M. P., and Kakinami, Y.: Global S4 index variations ob- References served using FORMOSAT-3/COSMIC GPS RO technique dur- ing a solar minimum year, J. Geophys. Res.-Space Physics, 117, Anthes, R. A.: Exploring Earth’s atmosphere with radio occulta- A09322, https://doi.org/10.1029/2012ja017966, 2012. tion: contributions to weather, climate and space weather, Atmos. Cai, Y., Bai, W., Wang, X., Sun, Y., Du, Q., Zhao, D., Meng, X., Meas. Tech., 4, 1077–1103, https://doi.org/10.5194/amt-4-1077- Liu, C., Xia, J., Wang, D., Wu, D., Li, W., Wu, C., and Liu, 2011, 2011. C.: In-orbit performance of GNOS on-board FY3-C and the en- Anthes, R. A., Rocken, C., and Kuo, Y.-H.: Applications of COS- hancements for FY3-D satellite, Adv. Space Res., 60, 2812– MIC to meteorology and climate, Terr. Atmos. Ocean. Sci., 11, 2821, https://doi.org/10.1016/j.asr.2017.05.001, 2017. 115–156, 2000. Cucurull, L. and Derber, J. C.: Operational implementa- Anthes, R. A., Bernhardt, P. A., Chen, Y., Cucurull, L., Dymond, tion of COSMIC observations into NCEP’s global data K. F., Ector, D., Healy, S. B., Ho, S.-P., Hunt, D. C., Kuo, assimilation system, Weather Forecast., 23, 702–711, Y.-H., Liu, H., Manning, K., McCormick, C., Meehan, T. K., https://doi.org/10.1175/2008WAF2007070.1, 2008. Randel, W. J., Rocken, C., Schreiner, W. S., Sokolovskiy, S. Culverwell, I. D., Lewis, H. W., Offiler, D., Marquardt, C., and Bur- V., Syndergaard, S., Thompson, D. C., Trenberth, K. E., Wee, rows, C. P.: The Radio Occultation Processing Package, ROPP, T.-K., Yen, N. L., and Zeng, Z.: The COSMIC/FORMOSAT- Atmos. Meas. Tech., 8, 1887–1899, https://doi.org/10.5194/amt- 3 mission: Early results, B. Am. Meteorol. Soc., 89, 313–333, 8-1887-2015, 2015. https://doi.org/10.1175/BAMS-89-3-313, 2008. Dach, R., Hugentobler, U., Fridez, P., and Meindl, M.: Bernese GPS Ao, C. O., Hajj, G. A., Meehan, T. K., Dong, D., Iijima, B. A., Software Version 5.0. Astronomical Institute, University of Bern, Mannucci, J. A., and Kursinski, E. R.: Rising and setting GPS Switzerland, 2007. occultations by use of open-loop tracking, J. Geophys. Res., 114, D04101, https://doi.org/10.1029/2008JD010483, 2009.

Atmos. Meas. Tech., 11, 5797–5811, 2018 www.atmos-meas-tech.net/11/5797/2018/ Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS 5809

Du, Q. F., Sun, Y. Q., Bai, W. H., Wang, X. Y., Wang, D. W., Meng, B., Ao, C., Wickert, J., Syndergaard, S., Lauritsen, K. X. G., Cai, Y. R., Liu, C. L., Wu, D., Wu, C. J., Li, W., Xia, J. B., Leroy, S., Kursinski, E. R., Kuo, Y-H., Foelsche, U., M., and Liu, C.: The Next Generation GNOS Instrument For FY- Schmidt, T., and Gorbunov, M.: Reproducibility of GPS ra- 3 Meteorological Satellites, IGARSS 2016, Beijing, 381–383, dio occultation data for climate monitoring: Profile-to-profile 2016. inter-comparison of CHAMP climate records 2002 to 2008 Edwards, P. G. and Pawlak, D.: Metop: The space segment for Eu- from six data centers, J. Geophys. Res., 117, D18111, metsat’s Polar System, ESA Bulletin, 102, 6–18, 2000. https://doi.org/10.1029/2012JD017665, 2012. Fjeldbo, G., Kliore, G. A., and Eshleman, V. R.: The neutral atmo- Huang, C.-Y., Kuo, Y.-H., Chen, S.-Y., Terng, C.-T., and Chien, sphere of Venus as studied with the Mariner V radio occultation F.-C., Lin, P.-L., Kueh, M.-T., Chen, S.-H., Yang, M.-J., Wang, experiments, Astron. J., 76, 123–140, 1971. C.-J., and Prasad Rao, A. S. K. A. V.: Impact of GPS radio oc- Foelsche, U., Pirscher, B., Borsche, M., Kirchengast, G., and cultation data assimilation on regional weather predictions, GPS Wickert, J.: Assessing the climate monitoring utility of Solut., 14, 35–49, https://doi.org/10.1007/s10291-009-0144-1, radio occultation data: From CHAMP to FORMOSAT- 2010. 3/COSMIC, Terr. Atmos. Ocean. Sci., 20, 155–170, Jakowski, N., Wehrenpfennig, A., Heise, S., Reigber, C., https://doi.org/10.3319/TAO.2008.01.14.01(F3C), 2009. Lühr, H., Grunwaldt, L., and Meehan, T. K.: GPS ra- Foelsche, U., Scherllin-Pirscher, B., Ladstädter, F., Steiner, A. dio occultation measurements of the ionosphere from K., and Kirchengast, G.: Refractivity and temperature cli- CHAMP: Early results, Geophys. Res. Lett., 29, 1457, mate records from multiple radio occultation satellites con- https://doi.org/10.1029/2001GL014364, 2002. sistent within 0.05%, Atmos. Meas. Tech., 4, 2007–2018, Kuo, Y.-H., Zou, X., Chen, S. J., Huang, W., and Guo, Y.-R., An- https://doi.org/10.5194/amt-4-2007-2011, 2011. thes, R. A., Exner, M., Hunt, D., Rocken, C., and Sokolovskiy, Gorbunov, M. E.: Ionospheric correction and statistical opti- S.: A GPS/MET sonding through an intense upper-level front, B. mization of radio occultation data, Radio Sci., 37, 1084, Am. Meteorol. Soc., 79, 617–626, 1998. https://doi.org/10.1029/2000RS002370, 2002. Kursinski, E. R., Hajj, G. A., Bertiger, W. I., Leroy, S. S., Mee- Gorbunov, M. E. and Lauritsen, K. B.: Analysis of wave han, T. K., Romans, L. J., Schofield, J. T., McCleese, D. J., Mel- fields by Fourier integral operators and their applica- bourne, W. G., Thornton, C. L., Yunck, T. P., Eyre, J. R., and tion for radio occultations, Radio Sci., 39, RS4010, Nagatani, R. N.: Initial Results of radio occultation observations https://doi.org/10.1029/2003RS002971, 2004. of Earth’s atmosphere using the Global Positioning System, Sci- Gorbunov, M. E., Gurvich, A. S., and Bengtsson, L.: Advanced al- ence, 271, 1107–1110, 1996. gorithms of inversion of GPS/MET satellite data and their appli- Kursinski, E. R., Hajj, G. A., Hardy, K. R., Schofield, J. T., and Lin- cation to reconstruction of temperature and humidity, Tech. Rep. field, R.: Observing Earth’s atmosphere with radio occultation 211, Max Planck Inst. for Meteorol., Hamburg, Germany, 1996. measurements, J. Geophys. Res., 102, 23429–23465, 1997. Gorbunov, M. E., Lauritsen, K. B., Benzon, H.-H., Larsen, G. Lackner, B. C., Steiner, A. K., Hegerl, G. C., and Kirchengast, B., Syndergaard, S., and Sørensen, M. B.: Processing of G.: Atmospheric climate change detection by radio occultation GRAS/METOP radio occultation data recorded in closed-loop data using a fingerprinting method, J. Climate, 24, 5275–5291, and raw-sampling modes, Atmos. Meas. Tech., 4, 1021–1026, https://doi.org/10.1175/2011JCLI3966.1, 2011. https://doi.org/10.5194/amt-4-1021-2011, 2011. Le Marshall, J., Xiao, Y., Norman, R., Zhang, K., Rea, A., Cucu- Harnisch, F., Healy, S. B., Bauer, P., and English, S. J.: Scaling rull, L., Seecamp, R., Steinle, P., Puri, K., and Le, T.: The benefi- of GNSS radio occultation impact with observation number us- cial impact of radio occultation observations on Australian region ing an ensemble of data assimilations, Mon. Weather Rev., 141, forecasts, Australian Meteorol. Oceanogr. J., 60, 121–125, 2010. 4395–4413, https://doi.org/10.1175/MWR-D-13-00098.1, 2013. Leroy, S. S.: The measurement of geopotential heights by GPS radio Healy, S. and Eyre, J. R.: Retrieving temperature, water vapor and occultation, J. Geophys. Res., 102, 6971–6986, 1997. surface pressure information from refractive index profiles de- Li, M., Li, W., Shi, C., Jiang, K., Guo, X., Dai, X., Meng, X., Yang, rived by radio occultation: A simulation study, Q. J. Roy. Mete- Z., Yang, G., and Liao, M.: Precise orbit determination of the orol. Soc., 126, 1661–1683, 2000. Fengyun-3C satellite using onboard GPS and BDS observations, Healy, S. B. and Thépaut, J.-N.: Assimilation experiments with J. Geodesy, 91, 1313–1327, 2017. CHAMP GPS radio occultation measurements, Q. J. Roy. Me- Liao, M., Zhang, P., Bi, Y. M., and Yang, G. L.: A preliminary teorol. Soc., 132, 605–623, 2006. estimation of the radio occultation products accuracy from the Ho, S.-P., Kirchengast, G., Leroy, S., Wickert, J., Mannucci, A. Fengyun-3C meteorological satellite, Acta Meteorol. Sin., 73, J., Steiner, A. K., Hunt, D., Schreiner, W., Sokolovskiy, S., 1131–1140, 2015. Ao, C., Borsche, M., von Engeln, A., Foelsche, U., Heise, S., Liao, M., Zhang, P., Yang, G. l., Bai, W. H., Meng, X. G., Du, Q. Iijima, B., Kuo, Y.-H., Kursinski, R., Pirscher, B., Ringer, M., F., and Sun, Y. Q.: Status of Radio Occultation Sounding Tech- Rocken, C., and Schmidt, T.: Estimating the uncertainty of us- nology of FY-3C GNOS, Adv. Meteorol. Sci. Technol., 6, 83–87, ing GPS radio occultation data for climate monitoring: Inter- 2016a. comparison of CHAMP refractivity climate records from 2002 to Liao, M., Zhang, P., Yang, G.-L., Bi, Y.-M., Liu, Y., Bai, W.- 2006 from different data centers, J. Geophys. Res., 114, D23107, H., Meng, X.-G., Du, Q.-F., and Sun, Y.-Q.: Preliminary https://doi.org/10.1029/2009JD011969, 2009. validation of the refractivity from the new radio occulta- Ho, S.-P., Hunt, D., Steiner, A. K., Mannucci, A. J., Kirchen- tion sounder GNOS/FY-3C, Atmos. Meas. Tech., 9, 781–792, gast, G., Gleisner, H., Heise, S., von Engeln, A., Mar- https://doi.org/10.5194/amt-9-781-2016, 2016b. quardt, C., Sokolovskiy, S., Schreiner, W., Scherllin-Pirscher,

www.atmos-meas-tech.net/11/5797/2018/ Atmos. Meas. Tech., 11, 5797–5811, 2018 5810 Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS

Liu, C. L., Kirchengast, G., Zhang, K. F., Norman, R., Li, wide GPS Network, Geophys. Res. Lett., 24, 2283–2286, Y., Zhang, S. C., Carter, B., Fritzer, J., Schwaerz, M., https://doi.org/10.1029/97gl02273, 1997. Choy, S. L., Wu, S. Q., and Tan, Z. X.: Characterisation Poli, P., Joiner, J., and Kursinski, E. R.: 1DVAR analy- of residual ionospheric errors in bending angles using GNSS sis of temperature and humidity using GPS radio oc- RO end-to-end simulations, Adv. Space Res., 52, 821–836, cultation refractivity data, J. Geophys. Res., 107, 4448, https://doi.org/10.1016/j.asr.2013.05.021, 2013. https://doi.org/10.1029/2001JD000935, 2002. Liu, C. L., Kirchengast, G., Zhang, K. F., Tan, Z. X., Johannes, F., Poli, P., Healy, S. B., Rabier, F., and Pailleux, J.: Preliminary as- and Sun, Y. Q.: The effects of residual ionospheric errors on GPS sessment of the scalability of GPS radio occultations impact in radio occultation temperature, Chinese J. Geophys., 57, 2404– numerical weather prediction, Geophys. Res. Lett., 35, L23811, 2414, 2014 (in Chinese). https://doi.org/10.1029/2008GL035873, 2008. Liu, C. L., Kirchengast, G., Zhang, K., Norman, R., Li, Y., Zhang, Rocken, C., Anthes, R., Exner, M., Hunt, D., Sokolovskiy, S., Ware, S. C., Fritzer, J., Schwaerz, M., Wu, S. Q., and Tan, Z. X.: R., Gorbunov, M., Schreiner, W., Feng, D., Herman, B., Kuo, Y.- Quantifying residual ionospheric errors in GNSS radio occul- H., and Zou, X.: Analysis and validation of GPS/MET data in the tation bending angles based on ensembles of profiles from neutral atmosphere, J. Geophys. Res., 102, 29849–29866, 1997. end-to-end simulations, Atmos. Meas. Tech., 8, 2999–3019, Rocken, C., Kuo, Y.-H., Schreiner, W., Hunt, D. C., and https://doi.org/10.5194/amt-8-2999-2015, 2015. Sokolovskiy, S. V.: COSMIC system description, Special Issue, Liu, C., Kirchengast, G., Sun, Y., Bai, W., Du, Q., Wang, Terr. Atmos. Ocean. Sci., 11, 21–52, 2000. X., Meng, X., Wang, D., Cai, Y., Wu, D., Wu, C., Li, Schmidt, T., Heise, S., Wickert, J., Beyerle, G., and Reigber, C.: W., Xia, J., and Liu, C.: Study of bending angle residual GPS radio occultation with CHAMP and SAC-C: global moni- ionosphric error in real RO data, 2016 IEEE International Geo- toring of thermal tropopause parameters, Atmos. Chem. Phys., 5, science and Remote Sensing Symposium (IGARSS), 4171– 1473–1488, https://doi.org/10.5194/acp-5-1473-2005, 2005. 4174, https://doi.org/10.1109/igarss.2016.7730087, 2016. Schmidt, T., Wickert, J., Beyerle, G., and Heise, S.: Liu, C., Kirchengast, G., Sun, Y., Zhang, K., Norman, R., Schwaerz, Global tropopause height trends estimated from GPS ra- M., Bai, W., Du, Q., and Li, Y.: Analysis of ionospheric struc- dio occultation data, Geophys. Res. Lett., 35, L11806, ture influences on residual ionospheric errors in GNSS radio oc- https://doi.org/10.1029/2008GL034012, 2008. cultation bending angles based on ray tracing simulations, At- Schmidt, T., Wickert, J., and Haser, A.: Variability of the upper tro- mos. Meas. Tech., 11, 2427–2440, https://doi.org/10.5194/amt- posphere and lower stratosphere observed with GPS radio oc- 11-2427-2018, 2018. cultation bending angles and temperatures, Adv. Space Res., 46, Liu, Y. and Xue, J.: Assimilation of GNSS radio occultation ob- 150–161, https://doi.org/10.1016/j.asr.2010.01.021, 2010. servations in GRAPES, Atmos. Meas. Tech., 7, 3935–3946, Schreiner, W., Rocken, C., Sokolovskiy, S., Syndergaard, S., and https://doi.org/10.5194/amt-7-3935-2014, 2014. Hunt, D.: Estimates of the precision of GPS radio occultations Loescher, A. and Kirchengast, G.: Variational data assimilation for from the COSMIC/FORMOSAT-3 mission, Geophys. Res. Lett., deriving global climate analyses from GNSS radio occultation 34, L04808, https://doi.org/10.1029/2006GL027557, 2007. data, GPS Solut., 12, 227–235, https://doi.org/10.1007/s10291- Sokolovskiy, S. V.: Tracking tropospheric radio occultation signals 008-0087-y, 2008. from low Earth orbit, Radio Sci., 36, 483–498, 2001. Loiselet, M., Stricker, N., Menard, Y., and Luntama, J.-P.: GRAS Sokolovskiy, S. V., Rocken, C., Lenschow, D. H., Kuo, Y.- – Metop’s GPS-based atmospheric sounder, ESA Bulletin, 102, H., Anthes,R. A., Schreiner, W. S., and Hunt, D. C.: Ob- 38–44, 2000. serving the moist troposphere with radio occultation sig- Luntama, J.-P., Kirchengast, G., Borsche, M., Foelsche, U., Steiner, nals from COSMIC, Geophys. Res. Lett., 34, L18802, A., Healy, S., von Engeln, A., O’Clerigh, E., and Marquardt, https://doi.org/10.1029/2007GL030458, 2007. C.: Prospects of the EPS GRAS mission for operational atmo- Sokolovskiy, S., Rocken, C., Schreiner, W., Hunt, D. C., and spheric applications, B. Am. Meteorol. Soc., 89, 1863–1875, Johnson, J.: Postprocessing of L1 GPS radio occultation sig- https://doi.org/10.1175/2008BAMS2399.1, 2008. nals recorded in open-loop mode, Radio Sci., 44, RS2002, Mao, T., Sun, L., Yang, G., Yue, X., Yu, T., Huang, C., Zeng, https://doi.org/10.1029/2008RS003907, 2009. Z., Wang, Y., and Wang, J.: First Ionospheric Radio-Occultation Steiner, A. K., Kirchengast, G., and Ladreiter, H. P.: Inversion, Measurements From GNSS Occultation Sounder on the Chinese error analysis, and validation of GPS/MET occultation data, Feng-Yun 3C Satellite, IEEE T. Geosci. Remote, 54, 5044–5053, Ann. Geophys., 17, 122–138, https://doi.org/10.1007/s00585- 2016. 999-0122-5, 1999. Melbourne, W. G., Davis, E. S., Duncan, C. B., Hajj, G. A., Hardy, Steiner, A. K., Kirchengast, G., Foelsche, U., Kornblueh, L., K. R., Kursinski, E. R., Meehan, T. K., Yong, L. E., and Yunck, Manzini, E., and Bengtsson, L., GNSS occultation sounding for T. P.: The application of spaceborne GPS to atmospheric limb climate monitoring, Phys. Chem. Earth (A), 26, 113–124, 2001. sounding and global change monitoring, JPL Publ. 94-18, Jet Steiner, A. K., Kirchengast, G., Lackner, B. C., Pirscher, B., Propulsion Lab, Calif. Inst. of Technol., Pasadena, CA, 1994. Borsche, M., and Foelsche, U.: Atmospheric temperature change Offiler, D.: The radio occultation processing package (ROPP) detection with GPS radio occultation 1995 to 2008, Geophys. an overview, Tech. Rep., GRAS SAF, Document-No: Res. Lett., 36, L18702, https://doi.org/10.1029/2009GL039777, SAF/GRAS/METO/UG/ROPP/001, 2008. 2009. Pi, X., Mannucci, A. J., Lindqwister, U. J., and Ho, C. M.: Mon- Steiner, A. K., Lackner, B. C., Ladstädter, F., Scherllin-Pirscher, B., itoring of global ionospheric irregularities using the World- Foelsche, U., and Kirchengast, G.: GPS radio occultation for cli-

Atmos. Meas. Tech., 11, 5797–5811, 2018 www.atmos-meas-tech.net/11/5797/2018/ Y. Sun et al.: The FengYun-3C radio occultation sounder GNOS 5811

mate monitoring and change detection, Radio Sci., 46, RS0D24, Ware, R., Exner, M., Feng, D., Gorbunov, M., Hardy, K., Her- https://doi.org/10.1029/2010RS004614, 2011. man, B., Kuo, Y., Meehan, T., Melbourne, W., Rocken, C., Steiner, A. K., Hunt, D., Ho, S.-P., Kirchengast, G., Mannucci, Schreiner, W., Sokolovskiy, S., Solheim, F., Zou, X., An- A. J., Scherllin-Pirscher, B., Gleisner, H., von Engeln, A., thes, R., Businger, S., and Trenberth, K.: GPS Sounding of Schmidt, T., Ao, C., Leroy, S. S., Kursinski, E. R., Foelsche, the atmosphere from Low Earth Orbit: Preliminary results, B. U., Gorbunov, M., Heise, S., Kuo, Y.-H., Lauritsen, K. B., Mar- Am. Meteorol. Soc., 77, 19–40, https://doi.org/10.1175/1520- quardt, C., Rocken, C., Schreiner, W., Sokolovskiy, S., Synder- 0477(1996)077<0019:GSOTAF>2.0.CO;2, 1996. gaard, S., and Wickert, J.: Quantification of structural uncer- Wickert, J.: Amplitude variations in GPS signals as a possible indi- tainty in climate data records from GPS radio occultation, At- cator of ionospheric structures, Geophys. Res. Lett., 31, L24801, mos. Chem. Phys., 13, 1469–1484, https://doi.org/10.5194/acp- https://doi.org/10.1029/2004gl020607, 2004. 13-1469-2013, 2013. Wickert, J., Reigber, C., Beyerle, G., Koenig, R., Marquardt, C., Sun Y. Q., Liu C. L., Du Q. F., Wang X. Y., Bai W. H., Kirchen- Schmidt, T., and Grunwaldt, L., Galas, R., Meehan, T. K., Mel- gast G., Xia J. M., Meng X. G., Wang D. W., Cai Y. R., bourne, W. G., and Hocke, K.: Atmosphere sounding by GPS ra- Zhao D. Y., Wu C. J., Li W., and Liu C.: Global Navigation dio occultation: First results from CHAMP, Geophys. Res. Lett., Satellite System Occultation Sounder II (GNOS II), IGARSS 28, 3263–3266, https://doi.org/10.1029/2001GL013117, 2001. 2017, Fort Worth, IEEE Xplore, 2017 IEEE International Geo- Wickert, J., Beyerle, G., Hajj, G. A., Schwieger, V., and science and Remote Sensing Symposium (IGARSS), 1189– Reigber, C.: GPS radio occultation with CHAMP: At- 1192, https://doi.org/10.1109/IGARSS.2017.8127170, 2017. mospheric profiling utilizing the space-based single dif- Syndergaard, S.: On the ionosphere calibration in GPS ra- ference technique, Geophys. Res. Lett., 29, 28-1–28-4, dio occultation measurements, Radio Sci., 35, 865–883, https://doi.org/10.1029/2001GL013982, 2002. https://doi.org/10.1029/1999rs002199, 2000. Wickert, J., Beyerle, G., König, R., Heise, S., Grunwaldt, L., Micha- Von Engeln, A., Healy, S., Marquardt, C., Andres, Y., lak, G., Reigber, Ch., and Schmidt, T.: GPS radio occultation and Sancho, F.: Validation of operational GRAS ra- with CHAMP and GRACE: A first look at a new and promis- dio occultation data, Geophys. Res. Lett., 36, L17809, ing satellite configuration for global atmospheric sounding, Ann. https://doi.org/10.1029/2009GL039968, 2009. Geophys., 23, 653–658, https://doi.org/10.5194/angeo-23-653- Vorob’ev, V. V. and Krasil’nikova, T. G.: Estimation of the accu- 2005, 2005. racy of the atmospheric refractive index recovery from doppler Wu, X., Hu, X., Gong, X., Zhang, X., and Wang, X.: Analysis of shift measurements at frequencies used in the NAVSTAR sys- inversion errors of ionospheric radio occultation, GPS Solutions, tem, Phys. Atmos. Ocean, 29, 602–609, 1994. 13, 231–239, https://doi.org/10.1007/s10291-008-0116-x, 2009. Wang, S. Z., Zhu, G. W., Bai, W. H., Liu, C. L., Sun, Y. Q., Du, Q. Xiong, C., Lu, C., Zhu, J., and Ding, H.: Orbit determination using F., Wang, X. Y., Meng, X. G., Yang, G. l., Yang, Z. D., Zhang, real tracking data from FY3C-GNOS, Adv. Space Res., 60, 543– X. X., Bi, Y. M., Wang, D. W., Xia, J. M., Wu, D., Cai, Y. R., 556, 2017. and Han, Y.: For the first time fengyun3 C satellite-global navi- Yang, G. L., Sun Y. Q., Bai, W. H., Zhang, X. X., Liu, C. L., Meng, gation satellite system occultation sounder achieved spaceborne X. G., Bi, Y. M., Wang, D. W., and Zhao, D. Y.: Validation results Bei Dou system radio occultation, Acta Phys. Sin., 64, 089301, of NmF2 and hmF2 derived from ionospheric density profiles 1–7, 2015. of GNOS on FY-3C Satellite, Science China Technological Sci- Wang, X. Y., Sun, Y. Q., Bai, W. H., Du, Q. F., Wang, D. W., Wu, ences, 59, 183–190, https://doi.org/10.1007/s11431-015-5920-2, D., Yu, Q. L., and Han, Y.: Simulation of number and distribution 2017. of Compass occultation events, Chinese J. Geophys., 56, 2521– Yue, X., Schreiner, W. S., Zeng, Z., Kuo, Y.-H., and Xue, 2530, 2013. X.: Case study on complex sporadic E layers observed by Wang, X. Y., Sun, Y. Q., Du, Q. F., Bai, W. H., Wang, D. W., Cai, Y. GPS radio occultations, Atmos. Meas. Tech., 8, 225–236, R., Wu, D., and Yu, Q. L.: GNOS – Radio Occultation Sounder https://doi.org/10.5194/amt-8-225-2015, 2015. on Board of Chinese FY3 Satellites, IGARSS 2014, Quebec, Zhao, Q., Wang, C., Guo, J., Yang, G., Liao, M., Ma, H., and 4982–4985, 2014. Liu, J.: Enhanced orbit determination for BeiDou satellites with Wang, X. Y., Sun, Y. Q., Du, Q. F., Wang, D. W., Wu, D., Cai, Y. R., FengYun-3C onboard GNSS data, GPS Solutions, 21, 1179– Wu, C. J., Bai, W. H., Xia, J. M., and Li, W.: An Integrated GNSS 1190, 2017. Remote Sensing Instrument and Its First GNSSR Airborne Ex- periment, IGARSS 2016, Beijing, 4827–4830, 2016.

www.atmos-meas-tech.net/11/5797/2018/ Atmos. Meas. Tech., 11, 5797–5811, 2018